摘要
介绍了一种利用阵列触觉传感器实现工件形状识别的新方法。该方法针对传感器采集的触觉图像的特殊性,在信号的预处理中采用Hopfield神经网络解决断线、畸变等问题。并以矩描绘子作为特征判别的主要依据,采用模板匹配法实现了工件形状的正确识别。
The paper introduce a kind of new method that realize the workpiece shape recognition based on tactile sensor. With a view to the particularity of tactile image, it adopt Hopfield Neural Network to resolve the problem of aberration and broken line. We get characters using Moment-Describe and realize accurate recognition using template matching.
出处
《机电工程》
CAS
2005年第9期43-45,共3页
Journal of Mechanical & Electrical Engineering
关键词
阵列触觉传感器
形状识别
神经网络
array tactile sensor
shape recognition
neural network
